Rhythmical analysis of the working from home playlists; is there correlation?

As the graph in slide 2 shows, the Maximum Concentration playlist contains 1 song with a valence level of 0.0000 and an energy level of 0.0181. This song appears to be underneath most energy levels of the rest of the songs in the playlists, which is around 0.035. This is the song Dovidenja by Darija Horvat (The outlier in the graph on Instrumental Study is actually the same song). The song is a piece for piano and very calm in the musical events in the song. On the left, you see Spotify’s attempt to show the tempo throughout the song. An important note to make is that this song is played in a time signature of 6/8. Listening to the song the tempogram is quite easy to interpret.

The song consists of a stable rhythmical part by both left and right hand. The differences between the dispersed tempo parts and the (more or les) stabilized at 100 BPM parts is not particularly clear when listening to the song. I suspect that the dispersion around 5 seconds starts because of a small, syncopated rhythm in the piano. I feel like the small part messes up the observation of the tempo by Spotify. I think this mess-up keeps on until 25 seconds. At the end of these 20 seconds it seems the song only plays half a bar, which results in another mess-up in the right direction; back to 100 BPM. Between 25 and 55 seconds the song becomes less rhythmical and near the 50 seconds it’s just chords. That’s why the tempo is so strongly defined at 100 BPM around those parts. Then, between 55 and 80 seconds the song is very syncopated again and a clear tempo can’t be determined. It feels in a way like a repetition of the first syncopated rhythm between 5 and 25 seconds. However i would need the score to know this for sure. In the last 5 seconds the tempogram shows a big blur with no clear tempo. This is because these last 5 seconds are the result of the piano slowing down in the song and sustaining the last few notes these last 5 seconds. During this sustained chord there is no such thing as rhythm, so it is quite logically that Spotify can’t find a proper tempo for this part.

What does seem particular about this song is that the Spotify features function labels this song with a tempo of 0.000 BPM. I think the high level of dispersion of the song and the different parts in the song result in a difficult to determine tempo and Spotify just states it as 0. It’s interesting to find out how the algorhythms behind Spotify actually work.

2020: What is the best music to be listening to, working from home?

2020 was the year of change for every one of us. Traveling has been brought back to a minimum, schools close and working from home became the standard. As we have all experienced, our computers developed into our community houses, classrooms, living rooms, pubs, clubs, cinemas, restaurants and what not. I believe it has brought us way more knowledge than we could have ever imagined. With the office disappearing out of peoples lives, a lot of regularities change as well. At home there are kids, distractions like hobby’s, TV, long overdue home-improvement projects and what not. This combination of impulses change working habits quite much. I wonder to what extent music-listenning changes, being one of those habits. I=Therefore i wonder how the music played working at homes differs from music played at the office. Luckily Spotify wouldn’t be Spotify if it didn’t turn every human habit into a playlist. When you search for working from home, you end up with hundreds and hundreds of playlists. I decided to limit this corpus to a special “working from home” folder with all playlists for working from home, created by Spotify. The playlists used are shown to the left. Categorization that the table shows has been added, to make analysis easier to execute and interpret. The categorization is based on the main functions that the playlists have been formed around or the kind of music it consists of. These are the main categorizations for Spotify playlists. Either a collection of kind of the same sounding music or a collection of music that supports you while doing something in particular. I found these categories to be the reason of existence for these playlists and therefore the main point of view to analyze the playlists

What does working from home mean for Spotify?

The graphs on the left show all the energy and valence levels according to Spotifys API. The playlists are categorized as stated in the previous page. What mainly becomes clear is that Spotify builds playlists for concentration around songs with low energy and low valence levels. In a cautious way you could say the same about the Jazz playlists. However, the songs are more scattered in the plot. The Lofi playlists ar generally low in energy, but more differing when it comes to valence. About the Work and Study playlists nothing really conclusive can be said. Maybe that the songs from To Do List are generally lower in Valence than the songs from Werkdag FM.

The “Lady” in the sheets: What makes Mujer an outlier in Jazz Vibes?

One particular outlier is the song Mujer by RLLBTS and Golden Ticket Tapes. This song has a valence of 0.9930 and energy level of 0.83400. Both rather high levels for the majority in the Jazz Vibes playlist. The graph on the right shows the chromagram of this particular song. In this visualization of the pitch classes in this rather short song, there are clear concentrations of energy on D, Eb, F and G. Besides these pitch classes, there is a lot of noise in all the other pitch classes. This chromagram can be better understood when the song is being listened to; the accents on the snare and hi-hats are probably the accents on F and G, since they are relatively stable sounds and are constant throughout the song. Also, around 60 seconds in to the song, some kind of under water-phaser effect is used. Here the drum hits are silent, which is clearly cisible in the chromagram. All the rubbish going on in the other pitch classes is a result of the samples used for this song. Multiple samples play at the same time during the song. These samples were either pulled from vinyl or processed to sound like they have been. Because of that, the chromagram shows a lot of dispersed energy among pitch classes. The samples are not clean. This does contribute to the song however, the identity of this song suggests an ambient-like, crackly atmosphere. Which is also something that could be said about the Lofi playlists. This is not very surprising, since one of the artists, Golden Ticket Tapes, has got songs in both the Lofi Hip-Hop and the Lo-Fi Beats playlist.

Why Lofi?

As we have seen in previous page, Lofi playlists have a particular sound, that Spotify thinks would help you with studying. I took an outlier from the Jazz Vibes playlist, which appeared to be more of a Lofi song, since it popped up in one of the Lofi playlists and the ambient, crackly sound identity was quite visible from the chromagram. This sound identity should also be visible from the timbre. That’s why i select an outlier from the Lofi graph, which is Sittin’ Around Smokin’ by sumishii, from the playlist Lo-Fi Beats. Question is whether or not this sound identity is becoming visible from the cepstrogram and/or self-similarity matrices. This cepstrogram shows quite a lot of energy in all of the timbre categories. This suggests a similar conclusion to the one the last page did; Lofi has an crackly, ambient like sound identity. Sounds are not clean, which results in a rumbly end product and can be seen in both chroma- and cepstrograms.

Sittin’ Around Smokin’: The song compared to itself

This page shows the Self-Similarity Matrix on Sittin’ Around Smokin’. The first thing that this graph shows are the parallel diagonal lines. Listening to the song, these similarity lines can be explained. The song is constructed out of samples, or a particular sample. This sample is used in repetition throughout the song, with small interruptions by vocal ad-lib like elements. The repetition of the sample results in the diagonal lines this graphs shows. Multiple elements of the song are completely the same throughout the song. This adds an element to our definition of Lofi; Ambient, crackly music, constructed of samples with a lot of repetition in it. The arrangements of Lofi songs are therefore rather simple. I suppose this simpleness is an important aspect for Spotify to adapt it in a study/work playlist folder. The repetitiveness and absence of extreme sounds suggest a calm listenning experience and could therefore improve cognitive abilities, according to Spotify.

The meaning of it all: wat works and what does not?

Spotify has once again succeeded in forming playlists with corresponding features. I wonder however whether or not the playlists contribute to a better environment while working or studying. Spotify clearly sees a couple of ways in which the playlists could help: The music is instrumental and low in energy(Calm & Focused, Deep Focus, Focus Modus, Instrumental Study and Maximum Concentration), or the music is a bunch of samples, crackling ambiently into your brain (Jazz Vibes, Lofi Beats, Lofi Hip-Hop, Lush Lofi), Jazz music in general(Coffe Table Jazz) or the music is just feel good pop songs that are easy on the ear and not to pungent in their arrangements(To Do List, Werkdag FM). Coming weeks i will further investigate whether any of these aspects contribute to concentration, study or work in general.

Away from the Lofi -> All playlists have timbre, how do they relate?

The graph on the left shows the dispersion of the timbre coefficients among the 4 categories of playlists. For computing reasons all playlists have been brought back to the first 30 songs of the playlist. The dispersion of the different playlists among the coefficients is relatively the same. However, the Concentration and Lofi categories have major dispersions in C02. The playlists from the Concentration and Lofi categories have proven to be correlative before. As the graph in the second slide showed, concerning energy and valence. The rest of this graph also shows that. The dispersions of the 2 categories are the most alike among the other playlists’ dispersions. This is clear through most of the other coefficients. I believe Spotify use moderately the same steps for forming Concentration playlists as for Lofi playlists. Soundwise they are both calm, low in volume and apparently easy on the timbre. I think these aspects of music could work for people listening to the music while working or studying. The arrangement of musical aspects in a playlist like these could result in less distraction while working, where jazz or pop-ish playlists, because of timbral tension, high energy levels and high loudness, could result in a higher level of distraction.